Between- and within-subject variation in nutrient intake from infancy to old age: estimating the number of days required to rank dietary intakes with desired precision

Estimating the accuracy of dietary questionnaire assessments: validation in terms of structural equation models

The validity and precision of questionnaire assessments of the habitual intake of individuals are usually evaluated by comparison with reference measurements that are supposed to provide a best possible substitute for the individuals' true intake values. In the present paper, a measurement error model is presented, defining different types of error--random or systematic, and within or between individuals--that may occur in dietary intake measurements. It is then discussed how simple latent variable models (structural equation models) can be used to estimate the average magnitude of these various types of error. So far, approaches described for the analysis of dietary validity studies have all been based on the assumption that the random errors of repeat reference measurements, taken by the same method on different occasions, are uncorrelated, so that the average of a sufficiently large number of repeat reference measurements will provide an accurate ranking of individuals by true intake level. In the present paper it is described how, by additional comparison with a third type of measurement such as a biochemical marker, the validity of dietary questionnaire measurements can be evaluated even in situations where the random errors of repeat reference measurements can no longer be assumed to be independent.

Estimating the variations and autocorrelations in dietary intakes on weekdays and weekends

The high incidence of breast cancer in the U.S. and the possible link with dietary fat has led to the development of educational programmes for reducing women's fat intakes by agencies such as the National Cancer Institute and the U.S. Department of Agriculture. In this paper, we analyse the effects of an intervention on the intakes of 12 nutrients by 37 women in the Houston area. We estimate a dynamic random effects model by maximum likelihood to estimate the between and the within variations and the autocorrelations using 7 consecutive food records before and after the intervention programme. The main findings are that the pattern of within variations differs during weekdays and weekends. Secondly, the mean intakes of nutrients such as beta-carotene and ascorbic acid tend to be lower on weekends. Lastly, the intervention programme reduced the overall fat intakes and also increased the variation in the consumption of foods high in fats during weekdays. We discuss the implications of the results for the design of further studies.

Diet quality index: capturing a multidimensional behavior

OBJECTIVE

Data for 5,484 adults (aged 21 years and older) who participated in the 1987-88 Nationwide Food Consumption Survey (NFCS) were used to develop an index of overall dietary intake that related to the major, diet-related, chronic diseases in the United States. The low response rate of the 1987-88 NFCS has raised concerns about potential bias, but this large data set is useful for methodologic studies and research that does not attempt to generalize the results to the US population.

ANALYSES

Dietary recommendations from the 1989 National Academy of Sciences publication Diet and Health were stratified into three levels of intake for scoring. Individuals who met a dietary goal were given a score of zero. Those who did not meet a goal, but had a fair diet, were given one point, and those who had a poor diet were given two points. These points were summed across eight diet variables to score the index from zero (excellent diet) to 16 (poor diet).

RESULTS

Lower index scores were positively associated with high intakes of other important measures of diet quality (eg, fiber, vitamin C). We found that single nutrients (such as dietary fat) were not necessarily associated with other measures of diet quality.

CONCLUSION

We concluded that this index ranking of overall dietary patterns was reflective of total diet quality, though substantial misclassification can result from using single nutrients or foods as indicators of diet quality.

Does diet provide adequate amounts of calcium, iron, magnesium, and zinc in a well-educated adult population?

Standard advice from dietitians, nutritionists, and physicians is that if one eats a well-balanced diet containing a variety of foods, supplements are not necessary. Little information is available, especially in those over 75, to determine whether actual diets do provide adequate amounts of these minerals. The participants of the Baltimore Longitudinal Study of Aging provide seven-day records which include vitamin and mineral supplement intakes. Median daily dietary intakes from diet in all 564 subjects and from diet plus supplements in those who use them were analyzed by age group and gender. More women than men took supplements. Median intakes of calcium from diet were below the recommended dietary allowance (RDA) for unsupplemented women and for supplemented women over 60. Approximately 25% of women under 50 and 10% of women over 50 consumed less than two thirds of the RDA for iron from diet. For both men and women, all groups had median diet intakes below the RDA for magnesium. Forty percent of men and about half of women consumed less than two thirds of the RDA. These results indicate that many people in this well-educated, presumably well-nourished population did not consume adequate amounts of calcium, iron, magnesium, and zinc from diet. More women than men are at risk. Even those taking supplements did not consume adequate levels of some minerals.

The degree of day-to-day variation in food intake in diabetic patients

A prospective study of food intake using 7 day food diaries was undertaken in 92 diabetic men and women aged 17-81 years. The median individual day-to-day coefficients of variation for energy intake were: in insulin treated patients 12.0%, in non-insulin treated patients 13.7%; for carbohydrate intake 14.5% and 13.8% and for fat 20.7% and 20.8%, respectively. The median individual differences between the minimum and maximum daily intake of energy in insulin treated patients was 787 kcal, in non-insulin treated patients 649 kcal, for carbohydrate intake 89g and 77g and fat 50g and 43g, respectively. Only 39% patients ate within 20% of their prescribed carbohydrate diet. In non-insulin treated patients on prescribed calorie controlled diets, calorie consumption was on average 46% in excess of that prescribed. Although the variation in dietary intake in diabetic patients is large, it is smaller than that reported in non-diabetic subjects in the UK. This variation is likely to make the manipulation of other antidiabetic therapy both difficult and somewhat arbitrary.

Validity of a short food frequency questionnaire for estimating nutrient intake in elderly people

Nutrient intakes estimated using a short self-administered semi-quantitative food frequency questionnaire (FFQ) were compared with results obtained from five 2 d diet records using household measures in a group of fifty-three elderly people (mean age 70 years) in Dunedin, New Zealand in 1989. Mean intakes for most nutrients were less than 5% different between the two methods. Correlations between the nutrient intake values (excluding supplements) from the diet records and those from the FFQ ranged from 0.34 for Zn in women to more than 0.75 for protein, Zn and Ca in men. For most nutrients, at least 70% of the subjects when classified by the food records fell into the same quintile or into the within-one-quintile category when classified by the FFQ. These data indicate that in elderly subjects a simple self-administered semi-quantitative FFQ can provide very similar information (for both group and individual intakes for many nutrients) to that obtained from 10 d of careful diet recording.

Vitamin C intake and plasma ascorbic acid concentration in adolescents

The relationship between vitamin C intake and status was investigated in a group of adolescents (13-14 years old). Dietary intakes were assessed using a 7 d weighted dietary record method, coupled with the collection of duplicate diets. Vitamin C intakes calculated using food composition tables were compared with values obtained by direct analysis of duplicate diets. Vitamin C status was judged via measurement of plasma ascorbic acid (AA) concentration in blood samples taken after a 12-15 h fast. The relationship between calculated and analysed vitamin C intake and plasma AA concentration was examined. Average daily calculated vitamin C intakes, for the group (n 54) as a whole over a 7 d period, gave a good estimate of intake, as judged by prompt analysis of duplicate diets. However, analysed v. calculated intakes were significantly different for approximately one-third of subjects when data were examined on an individual basis. Large discrepancies between analysed and calculated values could not be accounted for on a food group basis. In all but two individuals, calculated vitamin C intake was in excess of the new reference nutrient intake (RNI, part of the new daily reference values (Department of Health and Social Security, 1991)) of 40 mg and all plasma AA concentrations were well above those used to indicate even a moderate risk of deficiency. A relationship between vitamin C intake and plasma AA was observed for both males (n 19) and females (n 35). However, the relationship was much stronger for males who showed a wider range of both intake and plasma AA values.

Energy and nutrient intakes in congenital heart disease

Children with congenital heart disease may have significant growth retardation, which in part may be caused by insufficient dietary intake. Data on energy and nutrient intake were collected using a 14-day dietary record by weighing, in 22 children with congenital heart disease (mean age 39 months, range 12-126 months), prior to corrective operation. When viewed in relation to actual weight, energy intake averaged 88% (SD 17%) of that recommended by the FAO/WHO/UNU. Energy intakes and weight SD scores were significantly correlated (r = 0.55, p < 0.01). Protein intake was generally high, and even sufficient to allow catch-up growth. The majority of the children did not meet the recommendations for iron, zinc, calcium, or vitamins D, E, C, B1 or B6. Parents should be advised to give their children vitamin/mineral supplements, and to supply extra energy to children with failure to thrive.

Methodology and theory in human eating research

Current research in human eating is assessed from the perspective of current research methods which stress laboratory research, the use of artificial foods rather than real meals, shorter term studies, animal models, and abnormal eating models and an emphasis on sensory and physiological factors rather than social, cultural and contextual factors. The proposal is made to refocus more human eating research on real people eating real foods in real eating situations.

Estimating energy and nutrient intakes in studies of human fertility

Two methods of dietary recording, the 24-hr recall and the weighed dietary intake methods, are considered appropriate for estimating energy and nutrient intakes in studies of human fertility. The former method gives lower estimates than the latter, although weighed intakes may underestimate true intakes. Examination of food intakes of pregnant, lactating, and non-pregnant, non-lactating New Guinean women shows their diet to be less homogeneous than is generally assumed for groups in developing countries. As a result direct observations of food intake for a limited number of days are not sufficiently accurate for the estimation of intake of most of the nutrients examined. Rather the study design should reflect the variability of intakes of the nutrients and groups under consideration.

Inter- and intra-individual variability in food intake of elderly people in Perugia (Italy)

It has been shown that each individual has a considerable day-to-day variation (intra-individual variation) in his or her level of food consumption. A large intra-individual variation has adverse effects on the reliability of research studies. The effect of the intra-individual variation can be minimized by taking food intake records over several days. An increase in the number of days entails higher costs, and this could limit the practicability of the study. In the recent literature on the methodology of dietary surveys, there is a growing interest in the estimation of the number of days required to conduct a reliable dietary survey. Recent developments in statistical theory allow the problem of large intra-individual variability to be overcome. These new statistical techniques require knowledge of the intra- and inter-individual variability and the appropriate adjustment of the statistical results. The aim of the present study was to evaluate the number of days of recorded intake required to obtain an estimate of the components of variance (the inter-individual and the intra-individual variability) focusing on foods rather than nutrients.

Intake of intense sweeteners in Germany

The dietary intake of aspartame, cyclamate, and saccharin was evaluated in Germany (FRG) in 1988/89. In the first part of the study the sweetener intake was evaluated in a representative sample of the population. Complete 24-h records of the amount and type of all foods and drinks consumed were obtained from 2,291 individuals. The total daily intake was calculated for each person from the sweetener content of each product and was expressed in mg/kg body weight (bw). 35.9% of the participants ingested one or more sweeteners on the examination day. Cyclamate and saccharin were the prominent sweeteners because aspartame was at that time permitted only under special regulatory exemption, and products containing acesulfame were not yet available. For users of intense sweeteners the mean intakes of aspartame, cyclamate, and saccharin were 0.15, 2.62, and 0.250 mg/kg bw/day, respectively. At the 90th percentile of intake, i.e., for the heavy consumer, the ingestion of cyclamate and saccharin was about 2.5 times higher. Persons who adhered to a diet (diabetes, weight control) did not ingest sweeteners in substantially higher amounts. Tabletop sweeteners and beverages were the most important sources of sweeteners, and they contributed more than 80% of the total intake. Consumption of sweeteners in excess of the Acceptable Daily Intake (ADI) was rarely observed (saccharin: one person, cyclamate: 16 persons). In the second part of the study, the sweetener intake was further evaluated during a 7-day period in those subjects who in the 1-day study ingested any of the sweeteners in excess of 75% of the ADI. Complete 7-day food records were available from 40 out of the 41 subjects who fulfilled this criterium. In this selected subgroup in which 19 subjects were less than 19 years old, the mean daily intakes of aspartame, cyclamate, and saccharin were 0.13, 4.53, and 0.42 mg/kg body weight (bw), respectively. These levels correspond to 0.33, 41 and 17% of the corresponding ADI values. No subject exceeded the ADI of aspartame or saccharin on any day of the study. For cyclamate, the mean daily intake over the 7-day period exceeded the ADI in 4 subjects. The results indicate that at the time of the study the then valid German sweetener regulation protected the consumer adequately, and that the sweetener intake was in 99.8% of all examined persons within recommended limits.

Day-to-day variation in energy and nutrient intake: evidence of individuality in eating behaviour?

A study of within-subject variation in the reported food intakes and food selections of 29 adults participating in the Beltsville One-Year Dietary Intake Study suggests that individuals possess characteristics levels of day-to-day variation in intake. Application of Bartlett's test of homogeneity of variance indicated heterogeneity of within-subject variance in energy intake and in selected nutrients controlled for energy. Comparisons of estimates of within-subject variation derived from a split sample of days revealed the relative stability of individuals' magnitudes of day-to-day variation in energy and nutrient intake over time. For some nutrient variables, a large proportion of the observed heterogeneity in within-subject variation could be explained by a linear relationship between subjects' standard deviations and their mean intake levels. However, heterogeneity persisted when subjects' coefficients of variation were compared, suggesting that mean intake differences are insufficient to explain the observed differences in subjects' levels of day-to-day variation. The magnitude of observed variation in an individual's food intake and food selection from one day to the next appears to be a meaningful descriptor of that individual's intake behaviour.

Food use and health effects of soybean and sunflower oils

This review provides a scientific assessment of current knowledge of health effects of soybean oil (SBO) and sunflower oil (SFO). SBO and SFO both contain high levels of polyunsaturated fatty acids (PUFA) (60.8 and 69%, respectively), with a PUFA:saturated fat ratio of 4.0 for SBO and 6.4 for SFO. SFO contains 69% C18:2n-6 and less than 0.1% C18:3n-3, while SBO contains 54% C18:2n-6 and 7.2% C18:3n-3. Thus, SFO and SBO each provide adequate amounts of C18:2n-6, but of the two, SBO provides C18:3n-3 with a C18:2n-6:C18:3n-3 ratio of 7.1. Epidemiological evidence has suggested an inverse relationship between the consumption of diets high in vegetable fat and blood pressure, although clinical findings have been inconclusive. Recent dietary guidelines suggest the desirability of decreasing consumption of total and saturated fat and cholesterol, an objective that can be achieved by substituting such oils as SFO and SBO for animal fats. Such changes have consistently resulted in decreased total and low-density-lipoprotein cholesterol, which is thought to be favorable with respect to decreasing risk of cardiovascular disease. Also, decreases in high-density-lipoprotein cholesterol have raised some concern. Use of vegetable oils such as SFO and SBO increases C18:2n-6, decreases C20:4n-6, and slightly elevated C20:5n-3 and C22:6n-3 in platelets, changes that slightly inhibit platelet generation of thromboxane and ex vivo aggregation. Whether chronic use of these oils will effectively block thrombosis at sites of vascular injury, inhibit pathologic platelet vascular interactions associated with atherosclerosis, or reduce the incidence of acute vascular occlusion in the coronary or cerebral circulation is uncertain. Linoleic acid is needed for normal immune response, and essential fatty acid (EFA) deficiency impairs B and T cell-mediated responses. SBO and SFO can provide adequate linoleic acid for maintenance of the immune response. Excess linoleic acid has supported tumor growth in animals, an effect not verified by data from diverse human studies of risk, incidence, or progression of cancers of the breast and colon. Areas yet to be investigated include the differential effects of n-6- and n-3-containing oil on tumor development in humans and whether shorter-chain n-3 PUFA of plant origin such as found in SBO will modulate these actions of linoleic acid, as has been shown for the longer-chain n-3 PUFA of marine oils.(ABSTRACT TRUNCATED AT 400 WORDS)

A computerized diet questionnaire for use in diet health education. 1. Development and validation

A diet questionnaire was developed in association with a computer program to provide rapid nutritional feedback to the general public. The questionnaire was validated against 16 d of weighted diet records and biochemical variables in blood and urine. The highest Pearson correlation coefficients obtained between the questionnaire and the weighed records were for alcohol, fibre, iron, riboflavin (r 0.74, 0.67, 0.66, 0.66 respectively). Striking sex differences were shown in the results; the trend for higher correlations persisted in females. At least 65% of subjects were classified by questionnaire to within one quintile of the classification by weighed record for the majority of nutrients.

Minerals: exercise performance and supplementation in athletes

This paper examines whether mineral supplements are necessary for athletes, and whether these supplements will enhance performance. Macrominerals (calcium, magnesium, and phosphorus) and trace minerals (zinc, copper, selenium, chromium, and iron) are described. Calcium supplements are important for the health of bones. Athletes tend to have enhanced calcium status as assessed by bone mineral density, with the notable exception of female amenorrhoeic athletes. Magnesium status is adequate for most athletes, and there is no evidence that magnesium supplements can enhance performance. Phosphorus status is adequate for athletes. Phosphorus supplementation over an extended period of time can result in lowered blood calcium, however, some studies have shown that acute 'phosphate loading' will enhance performance. Athletes may have a zinc deficiency induced by poor diet and loss of zinc in sweat and urine. Limited data exist on the relationship of performance and zinc status. Widespread deficiencies in copper have not been documented, and there are no data to suggest that copper supplementation will enhance performance. There is no reason to suspect a selenium deficiency in athletes. The relationship between selenium status and performance has not been established, but selenium may play a role as an antioxidant. Because of the low intakes of chromium for the general population, there is a possibility that athletes may be deficient. Exercise may create a loss in chromium because of increased excretion into the urine. Many athletes, particularly female, are iron depleted, but true iron deficiencies are rare. Iron depletion does not affect exercise performance but iron deficiency anaemia does. Iron supplements have not been shown to enhance performance except where iron deficiency anaemia exists. In conclusion, poor diets are perhaps the main reason for any mineral deficiencies found in athletes, although in certain cases exercise could contribute to the deficiency. Mineral supplementation may be important to ensure good health, but few studies have definitively documented any beneficial effect of mineral supplementation on performance.

Nutrient intakes of different social-class groups: results from the Scottish Heart Health Study (SHHS)

Food frequency questionnaire and socio-demographic data were collected from over 10,000 Scottish men and women aged 40-59 years in a cross-sectional study of coronary heart disease (CHD) risk factors. Dietary intake, including the antioxidant vitamins C and E and beta-carotene, was assessed for different socio-economic groups. Trends in nutrient intakes were found with social-class (occupational) groups I-V. The non-manual-manual distinctions were clear even after standardizing for serum cotinine, and alternative classification by housing tenure and level of education did not confound the social-class effect. Total energy intake was significantly higher in the manual (men 10,363 KJ, women 7507 KJ) than in the non-manual (men 9156 KJ, women 7169 KJ) groups, and all nutrient amounts except for vitamin C, vitamin E, beta-carotene and fibre were significantly higher in the manual than the non-manual groups. Alcohol intake was lower in manual women, but higher in manual men compared with their respective non-manual groups. Sex and social-class differences were maintained after adjusting for total energy. Women in general, and manual women in particular, had the highest percentage energy from total fat (40.2) and saturated fat (18.2), while the percentage energy from polyunsaturated fat was lower in men than women, and lowest in manual men (4.4). The polyunsaturated:saturated fat (P:S) ratios were, for non-manual and manual men 0.32 and 0.31, and for non-manual and manual women 0.31 and 0.28. Fibre and antioxidant vitamin intakes, when expressed as nutrient densities, were lower in men than women, and lowest in manual men. Overall, men and women in manual occupations had a poorer-quality diet than did those in non-manual occupations. The coincident low P:S ratios and low antioxidant vitamin intakes in manual groups may contribute to an increased risk of CHD. Thus, the findings are compatible with the view that poor diet may be a contributory factor to the higher mortality rates for CHD which occur in the lower socio-economic groups.

Antioxidant vitamin intakes assessed using a food-frequency questionnaire: correlation with biochemical status in smokers and non-smokers

The increasing interest in the possible role of antioxidant vitamins in many disease states means that methods of assessing vitamin intakes which are suitable for large-scale investigations are now required. The suitability of the food-frequency questionnaire, which was developed by the Medical Research Council - Cardiff Group, for determining dietary intake of antioxidant vitamins in epidemiological studies was investigated in 196 Scottish men. The validity of the dietary data was assessed by comparison with serum vitamin concentrations, and separate analyses were performed for current smokers and non-smokers. The results showed that total energy intake and the percentage of energy derived from sugar were higher in smokers, and that both dietary and serum values of vitamin C, beta-carotene and vitamin E were lower in smokers than non-smokers. After adjustment for serum lipids, energy intake and body mass index, correlation coefficients between dietary and serum vitamins C and E were similar for smokers (r 0.555 and 0.25 respectively) and non-smokers (r 0.58 and 0.32 respectively). Correlation between dietary and serum carotenes was reduced from 0.28 in non-smokers to 0.09 in smokers and correlations for retinol and total vitamin A were weakly significant only for non-smokers. The food-frequency questionnaire assigned greater than 70% of subjects correctly into the upper or lower plus adjacent tertiles of serum vitamin values, with the exception of beta-carotene and total vitamin A for smokers. Thus, the food-frequency questionnaire appeared to be an adequate tool for assigning individuals into tertiles of serum antioxidant vitamins with the main exception of beta-carotene for smokers. Marked differences do occur between the vitamins and between the smoking groups which may reflect reduced accuracy of reporting on the food-frequency questionnaire or differential absorption and metabolism of the vitamins.

Nutritional assessment of children with juvenile chronic arthritis

Nutritional assessment was performed in 38 children with juvenile chronic arthritis (polyarticular [15]; pauciarticular [13]; systemic [10]). This included anthropometric measurements (height, weight, mid-upper arm circumference and four skinfolds) and a 7 day weighed food record. The children were classified according to current type of arthritis which in some cases was not the same as the initial onset type. Mean height Z scores were significantly below the Z score population mean of 0 in the systemic (P = 0.02) and polyarticular (P = 0.009) groups. Mean weight Z scores were also below the mean in the polyarticular group (P = 0.001) but the systemic group did not reach significance. Mean energy intakes were significantly below the recommended dietary intake (RDI) in the systemic (P = 0.01) and polyarticular (P = 0.001) groups. Mean intakes of calcium and zinc were below the RDI of 100% in the polyarticular group (P = 0.0001). Thirteen (34%) of children were taking some form of self-prescribed vitamin and/or mineral supplement. Vitamin C was the most commonly used supplement, despite dietary intakes greatly exceeding the RDI. Nutritional assessment is essential in the management of patients with JCA, in particular those with systemic or polyarticular disease.

Calcium intake and bone mass: a quantitative review of the evidence

The relationship between calcium intake and bone mass remains controversial. In this paper, the published research on this association is reviewed using the quantitative technique of meta-analysis. Selection of studies was based on defined eligibility criteria, and information relating to study design was recorded. Study results were converted, where necessary, to similar outcome measures so that direct comparison among studies was possible. A total of 37 eligible papers, representing 49 separate studies or parts of studies, were identified in the literature. Calcium had a consistent prevention effect on the rate of bone loss in the 12 studies of calcium supplements in postmenopausal women. This effect was greatest in studies in which the baseline calcium was low, supporting the idea of a threshold beyond which the effect of calcium is reduced. Cross-sectional studies showed a small but consistent positive correlation between calcium intake and bone mass. This association was greater in studies of premenopausal women. Some caution is needed in interpreting the results of this meta-analysis because of the poor quality of many of the studies reviewed. Nevertheless, the consistency of findings suggests that women in their early postmenopausal years will benefit from a high calcium intake.